Adaptyv Bio
Lausanne protein-design validation lab-as-a-service — runs open binder-design competitions and experimentally validates AI-designed proteins.
1. Core Product / Service
Adaptyv Bio (Adaptyv Biosystems) runs an automated protein-validation wet lab in Lausanne: customers submit protein designs as digital sequences via web platform or API, and Adaptyv synthesizes the DNA, expresses proteins cell-free, and measures binding (e.g. via SPR), returning experimental data [1]. Assays cover automated expression and purification plus BLI/SPR binding kinetics; it does not do structure determination [5]. The company is the experimental counterpart to computational protein design — it closes the loop between AI-generated designs and physical validation.
Its second product is open protein-design competitions, published on the Proteinbase hub (launched Oct 6, 2025), where designs are experimentally validated in Adaptyv's lab and results are released publicly [2]. Reported runs: EGFR binder competitions (201 designs / 2.5% hit rate → 400 designs / 14% hit rate, best 1.21 nM), a Nipah binder competition (1,200 tested, best 0.37 nM), and a May 2026 TREM2 hackathon where human teams (38.5% hit rate) beat autonomous LLM agents (34.3%) [2]. An Anthropic × Adaptyv Protein Design Competition (Sep–Dec 2026, five challenges, 5,000+ designs) is the headline 2026 partnership [2].
2. Target Users & Pain Points
- AI protein-design teams (academic and startup) that can generate candidate binders but cannot wet-lab validate them — Adaptyv is the validation layer.
- Pharma / biotech evaluating whether computationally designed binders are real before committing to a program.
Pain solved: the wet-lab validation step is the bottleneck of the design loop. Adaptyv turns "does my AI-designed protein actually bind?" into an API call, making validation a commodity rather than an in-house capital expense. Its automated-lab equipment investment is estimated at roughly $2–5M (Thermo Fisher / PerkinElmer as main vendors) — on the order of a tenth of total funding, with people, DNA consumables and platform R&D the larger costs [1][5].
Pricing: public list prices — from ~$149 per protein design (expression), ~$169 with binding characterization, expression-only from $49; 60-design minimum; ~20% discount for releasing results to Proteinbase. The entry price was previously $99, so pricing has been rising [5].
Customers: named customers include Anthropic, Google DeepMind, chai-discovery, Boltz, Roche, Novo Nordisk and Microsoft; the company cites 100+ customers and ~5× annual throughput growth [5].
3. Competitive Landscape
| Player | Type | Notes |
|---|---|---|
| Adaptyv Bio | Validation lab + open competitions | Proteinbase benchmark leader |
| Twist Bioscience | DNA synthesis | Upstream synthesis scale |
| Ginkgo Bioworks | Synbio platform | Broader foundry, less protein-design-focused |
| generate-biomedicines / evolutionary-scale | AI design | Compute-side design, not validation |
| lila-sciences | Autonomous lab | Closed-loop discovery, no public benchmark |
Adaptyv differentiates through the open, public benchmark format — the competitions themselves became an industry reference because results are openly comparable [1][2].
4. Unique Observations
- Open competitions became the industry benchmark. Adaptyv's public, experimentally-scored binder challenges (hit rates, affinities published) turned an automated lab into a neutral evaluation ground — the protein-design equivalent of a benchmark leaderboard. That is the main reason third parties (Anthropic, EPFL, GEM) run their evaluations through it [1][2].
- The "verification" middle layer. Between AI design (alphafold3, RFdiffusion-style tools, david-baker's lab) and manufacturing, Adaptyv occupies a niche that's neither compute nor drug development — a pure measurement utility. It fits the broader "verify–predict–optimize" closed loop that wet-lab automation unlocks [1].
- Price anchor for binder validation. Among binder-validation providers, Adaptyv is the one with public per-design pricing ($149–169) plus benchmark/competition infrastructure, effectively setting the reference price; the move up from $99 suggests tight capacity or deliberate customer selection [5].
- Human vs agent gap is measurable. The TREM2 hackathon result (humans 38.5% vs autonomous agents 34.3%) is an early, concrete data point on how far autonomous protein-design agents still trail expert humans [2].
5. Financials / Funding
- YC S21 (Y Combinator) [3].
- CHF 2.5M from Swiss VC Wingman Ventures (pre-seed) [3].
- $8M Seed (Sep 2025), led by ACE Ventures, with byFounders, Founderful, LongGame and angel investors [2].
- $40M Series A, led by Highland Europe, with ACE Ventures doubling down plus byFounders and Y Combinator — to triple lab capacity by end-2026, open a London lab/office in Q4 2026, and grow the team from ~25 to ~60 [2].
- Total raised: ≈ $51M (pre-seed/YC + $8M seed + $40M Series A), superseding the ~$11.17M figure CB Insights showed before the Series A [4][5].
- Founded: 2020 (company registered 2021) [5].
6. People & Relationships
- CEO / co-founder: Julian Englert.
- CTO / co-founder: Daniel Nakhaee-Zadeh Gutierrez.
- Co-founders: Moustafa Houmani, Amir Shahein.
- Customers: Anthropic, Google DeepMind, chai-discovery, Boltz, Roche, Novo Nordisk, Microsoft [5].
- Origin: founded 2020 (registered 2021); EPFL-affiliated team, Lausanne; co-organizes the EPFL Protein Design Week Myostatin binder challenge [2][3].
- Partners: Anthropic (2026 competition), EPFL, GEM (RBX1 competition), muni (TREM2 hackathon) [2].
- Competitors / adjacent: Twist Bioscience, Ginkgo Bioworks, generate-biomedicines, evolutionary-scale, lila-sciences.
Sources
[1] Internal research notes — Adaptyv vs Twist comparison; why the open competition became an industry benchmark; binder challenge mechanics; automated-lab equipment (~$2–5M); Anthropic collaboration. [2] WOWTALE, "Adaptyv Bio Raises $40M to Automate the Lab That Validates AI-Designed Proteins," https://en.wowtale.net/2026/08/29/234934/ (2026-10-05); Proteinbase, "EPFL Protein Design Week x Adaptyv," https://proteinbase.com/competitions/epfl-adaptyv-myostatin (2026-10-05) [3] EPFL News, "Adaptyv Biosystems secures funding from Y Combinator," https://news.epfl.ch/news/adaptyv-biosystems-secures-funding-from-y-combinat/ (2026-10-05) [4] CB Insights, "Adaptyv Biosystems," https://www.cbinsights.com/company/adaptyv-biosystems (2026-10-05) [5] Internal research notes: RNA/protein CRO lab directory; binder-validation provider comparison (Adaptyv vs Twist vs Ginkgo Datapoints, 2026-10-04); Adaptyv lab-equipment cost estimate — founding date, ~$51.2M cumulative funding, public pricing, named customers, equipment share of funding.
Adaptyv Bio
按设计公开标价的蛋白 binder 验证服务,运营 Proteinbase 与竞赛
- 数据类型
- 蛋白 binder 结合测定;表达量;热稳定性
- 规模口径
- EGFR 竞赛 >1,800 设计提交、601 个表征
- 客户
- Anthropic;Roche;Novo Nordisk
做什么
瑞士的蛋白湿实验验证公司:客户提交序列,返回表达与结合测定结果。官网现行标价为起价 $149/蛋白,含 2 次重复,标准周期 3 周;最快 3 天的 Prime 档为私测 [1]。早期报道价 $99–169 仅作历史参考 ⚠️;曾公布的 $49 表达 / $99 结合分级价,官网本次抓取未渲染,/pricing 页已 404。
数据怎么生产、谁付费
无细胞表达加结合测定,另有表达、热稳定性项目;检测方法按档位不同(Anthropic 案例页为 SPR,5 浓度、双重复 [2]),不宜笼统写成 BLI。付费方为模型公司与药企(公司口径 100+ 客户,含 Roche、Novo Nordisk [3])。公开结果可进入 proteinbase,含阴性结果。
谁在用、用在哪一段
anthropic-binder-study 用其验证 1,320 个设计、354 个结合(26.8%)[2];自办 EGFR 等公开竞赛(见 adaptyv-competition、benchbb)。EGFR 竞赛 >1,800 设计提交、601 个表征 [4]。属评测与闭环验证段,不是预训练数据来源。
争议与口径
融资:种子 $8M,2026-08 下旬 A 轮 $40M(Highland Europe 领投),累计约 $4,800 万,日期各报道不一 [3]。EGFR round 1 命中率有 14%(53/378 已表达)与约 3% 两种分母 [4]。公司未披露毛利与客户明细。
- https://www.adaptyvbio.com (2026-10-15)
- https://adaptyvbio.substack.com/p/case-study-benchmarking-claudes-protein (2026-10-15)
- https://aiunderstanding.org/news/wowtale-reports-adaptyv-bio-raised-40-million-to-automate-validation-of-ai-designe (2026-10-15)
- https://www.biorxiv.org/content/10.1101/2025.04.17.648362.full.pdf (2026-10-15)